Traditional deduplication systems typically utilize an in-memory hook table and store fingerprints on-disk in batches for comparison when backing up data. However, for large amounts of data, the size of the hook table may become large, the sampling rate of the hook table may be low, the necessary computing resources to perform deduplication may increase, and the deduplication capacity of the system may be reduced. In view of the foregoing, it may be understood that there may be significant problems and shortcomings associated with traditional deduplication technologies.